In an advanced information society where everything is connected to the network, such as IoE (Internet of Everything), a goal is to capture signs before an unexpected situation occurs and eventually maintain a normal state, by digitizing a state of everything into a numerical value from the health management of people or pets to the maintenance and management of production equipment or household appliances and preservation of the natural environment, and by analyzing the digitized value by a computer.
In order to realize the object, a system for monitoring the state of various items is an important element, and the development of a diagnosis system which uses various sensors (image sensors, vibration sensors, or the like) has proceeded. For example, in a production site or a production plant, transition to a diagnosis system for predicting the occurrence of a failure based on state monitoring data output from a sensor from a management method of the related art for periodically inspecting an operating pump, a compressor, or the like, is tested, and reduction of maintenance and management costs and deterrence of loss due to an unexpected failure, are expected.
In this type of diagnosis system, there are cases where maintenance and management are optimized by capturing signs of failures based on a difference between a numerical value of temperature or vibration state observed from a device which is a monitoring target and a numerical value in a normal state (for example, PTL 1).